No better method having been developed, in democratic republics like the United States elections are used to determine the distribution of immense power and the control of the disposal of huge funds. Fraud has thus been an ever-present companion. Tracy Campbell provides one of the best summaries of that problem in his book Deliver The Vote: A History Of Election Fraud, An American Political Tradition 1742-2004. Prof. Doug Jones Brief Illustrated History of Voting is also recommended.
Once upon a time, say around 1970, after a couple of centuries, some might say millennia, of experimenting methods had been developed to make election fraud quite difficult. Counties were divided into precincts, parishes, or townships with usually 1,000-2,000 registered voters in each precinct. Sometime after becoming eligible to vote in the precinct, potential electors went to the county clerk's office and provided proof that they were citizens and had physically resided in the precinct for the requisite period of time.
On the day of an election they then went to their polling place that was run by citizen election judges from both major parties and usually residents of that precinct. Upon identifying themselves they were required to sign the poll book compiled from all voters who were registered in that precinct. Since many of the voters and judges knew one another there was a substantial risk that an interloper would be recognized and challenged. And additional volunteer poll watchers were frequently present to further ensure honesty.
If a voter was physically unable to go to their polling place on Election Day they could request an absentee ballot from the county clerk. But the reasons a voter might use to obtain an absentee ballot were limited and usually the voter had to request the absentee ballot as well as deliver it to the clerk in person for deposit into a ballot box. Thus, very few absentee ballots were issued or cast.
Once the elector signed the poll book they were given a paper ballot and provided a private booth in which they could mark the ballot by hand as they chose without fear of reprisal or intimidation. No campaigning or electioneering was allowed in or around the polling place. Once they had marked the ballot as they freely chose any identifying marks or stubs used for ballot inventory were removed and the ballot was deposited in a sealed box that the election judges had verified was empty before the polls opened. The voter was not allowed to leave the polling place with their ballot in hand or to take pictures of it in order to forestall vote buying and selling.
Experience showed that strange things happened to ballots and ballot boxes once they left the polling place. So when the polls closed the election judges opened the ballot box and jointly hand counted the results, a process also open to citizen observers. Once the ballots were tallied a public notice of the results was prominently posted at the polling place. The ballots were returned to the ballot box and sealed. The sealed ballot box and the precinct results were then transported to the county clerk either by sworn peace officers or by at least two election judges, one from each major party. The county clerk then compiled the results from each precinct and published the election results.
In the main, elections are run by county clerks who don't know an electron from a proton, and snake oil salesmen and women found a new calling selling electronic voting machines to these clerks in the 1990's. There were, and are no limits to the promises made by vendors of these voting machines. And few county clerks and other election officials were, or are willing to admit mistakes, malfeasance, fraud, incompetence, and vendors often willingly lie to cover up the numerous problems, i.e., "glitches" in their vernacular, with their equipment. And county clerks and election officials frequently step into well-paid positions with these election companies when they leave public service after spending millions of taxpayer dollars on the vendor's voting machines.
After the disastrous presidential election in 2000, when the U.S. Supreme Court decided it somehow had the authority to tell states when to stop counting votes and thereby determine who became president, the next obvious step was for Congress to get into the act.
Diebold, headquartered in Ohio, had decided that voting machines were going to be profitable and together with their pet congressman, Bob Ney (R-Ohio), and lobbyist Jack Abramoff, pushed through what is known as the Help America Vote Act (HAVA). Ney and Abramoff were later convicted and jailed for a variety of nefarious acts but HAVA provided some $4 billion to "modernize" elections with electronic voting machines from vendors like Diebold, ES&S, Hart Intercivic, Sequoia, etc.
The Equal Justice Foundation, among many others, has extensively, but not exhaustively documented the disasters resulting from the widespread adoption of electronic voting machines.
Because county voter rolls were commonly disasters, often with more registered voters than residents, HAVA also mandated statewide voter registration databases. That, in turn, led to even more problems. At one point after implementing a statewide voter database, years late and way over budget, Colorado had about 5.4 million "registered voters" but only roughly 3.5 million residents eligible to vote.
HAVA also mandated at least one voting machine per precinct that allowed the disabled to vote without assistance. Since there are 185,000-190,000 precincts in the United States, and each such machine costs $4,000-$5,000, $4 billion was hardly sufficient and the states and counties were to pay the remainder. These direct recording electronic (DRE) voting machines proved to be a disaster in practice with breakdowns, long lines, lost votes, and on and on. However, voting machine vendors offered an alternative optical scanner that, optimistically (pun intended), reads the ballot and reports how the elector voted. As Arthur C. Clarke famously stated "Any technology sufficiently advanced is indistinguishable from magic." And to county clerks these magical machines seemed to work wonderfully. They also had the advantage that voters had no access to the inner workings which were all-too-often readily apparent with DREs. Optical scanners could also "read" thousands of ballots very quickly and very few were needed compared to DREs.
Not long after HAVA was passed county clerks began pushing the idea of using mail ballots, or "no excuse" absentee voting in order to conduct elections. By 2006 it was obvious that:
But once started modernizing elections, and with billions in public funding, why not try some other ways to screw up elections using computers? Space doesn't allow a complete tabulation and readers are referred to groups like Bev Harris' Black Box Voting, VotersUnite.org, and many, many others for details on the disastrous state of elections in general. Here I want to look at just a few of the problems the widespread switch to "no excuse," or total mail-ballot elections has cost citizens in further loss of election integrity.
The move to mail ballots dominates the west (see map) and Florida, of course. Oregon and Washington have now switched to all mail elections all the time. Other states, like Colorado, use all mail elections in municipal, local, and off-year elections. But even in federal elections many states now permit, nay, encourage the use of "no excuse" absentee, i.e., mail ballots.
Bear in mind that the following tabulation is a bare and incomplete outline of just a few of the known problems with absentee/mail ballots and why they were not used or allowed in the prior millennium.
Most state constitutions now guarantee the right of a voter to cast a ballot that has no identifying marks that would allow anyone to trace the ballot back to the individual who cast it. The conundrum of positively identifying a voter and yet ensuring their ballot and how they voted is secret is easily solved at a polling place but impossible with mail ballots.
In many places the voter is sent a ballot with a detachable stub at the top that identifies the ballot either with a number or bar code so the clerk can verify the ballot sent out to a voter is the same ballot that was returned. The voter is also instructed to sign the envelope before returning it and election officials now typically use machines to compare the signature on the envelope with the voter's signature on the registration form. All this flim-flam makes it look like the clerk is concerned with security but, in fact, simply insures that any insider, post office worker, or anyone else who handles the ballots can determine how an individual voted. And that has been demonstrated in several Colorado counties.
In other cases ballots to one group have been placed in different colored or sized envelopes, or the voter's party affiliation stamped on the outside of the envelope. Bar codes are common as well and there are many apps available to read these codes.
Even worse, some systems like Hart Intercivic place identifying marks, either bar codes or numbers, that tie the ballot directly to the voter even after the stub at the top of the ballot is removed. Of course that is true with these machines whether or not the ballot is mailed or voted in a polling place.
Detecting election fraud or simple incompetence has always depended heavily on reference to signed poll books to establish that the voter is corporeal (not only the dead vote by mail but ghosts as well), still lives in the precinct (hard to vote in multiple states or from a foreign country at a precinct), and that the voter only votes once (many cases of voters receiving multiple mail ballots from multiple locations have been documented).
Since the move to statewide voter registration it is also common, if not the rule, to register to vote by mail. So it is easy enough for an enterprising individual or group to fill out multiple voter registration forms and mail them in. A classic case of this occurred in South Dakota in 2002. If one is a bit more intelligent than that individual and puts the registration forms in separate envelopes with different return addresses and mails them on different days, detection is unlikely. Once registered they can then vote all those ghosts by mail in every election. And at least in Colorado a registered voter can list up to three addresses for a mail ballot, and there is no reason the address to send the ballot to can't be a mail forwarding agency.
And a poll book today is little more than a check in a database stating a ballot was returned from such and such an entity, corporeal or not, resident of the precinct, county, state, or even country or not.
Long experience taught that no candidate or proponent for an issue should be allowed any contact with the voter while they have their ballot in hand. Even today polling places put out signs around the area stating No Electioneering within a specified number of feet. Nor is anyone allowed to wear or bring campaign material into the polling place, or talk politics. Generally, even campaign buttons, T-shirts, or other such paraphernalia are not allowed.
Forget that antiquated practice with mail ballots. Usually, and by law, county clerks make available voter lists indicating who receives a mail ballot. Everyone, of course, in an all-mail election just to take any guess work out. The date ballots are mailed is public information and three days later the electioneering begins in earnest with innumerable phone calls, emails, personal visits, rallies, labor union or employee meetings at companies, party hacks, relations, etc., all pressuring electors to vote the party line, their conscience and good sense be damned.
Vote buying and selling has always been a popular way to rig American elections. To be successful the person who buys or sells a vote has to prove the ballot was voted the way the purchaser wants it to be. At a polling place considerable precautions are taken to insure that no one sees how an elector votes, that a ballot is not given to a voter until they have signed the poll book, that no copy of the elector's voted ballot is made, and that the ballot never leaves the polling place.
All those precautions go out the window with mail ballots. The ballot is uncontrolled in the hands of voters for days or weeks, giving them plenty of time to arrange to buy or sell their vote. The buyer can easily check the ballot was voted they want, and after the ballot is signed and sealed, drop it in a mail box anonymously to be sure the seller doesn't simply back out of the deal.
One of the basic problems with committing vote fraud is obtaining a ballot. As a result election officials traditionally were very careful to keep a close inventory of all ballots, from the number printed, to ballots issued to voters, spoiled ballots, and unused ballots remaining at the end of the election. Obviously the number of voted, spoiled, and remaining should add up to the number of ballots printed.
However, maintaining even a semblance of a ballot inventory becomes impossible with mail ballots. Depending on the election somewhere between 10% and 90% of the ballots mailed out will not be returned and commonly there are thousands of ballots floating around. So there is no problem obtaining a ballot. Clerks go to considerable effort to ensure the ballot returned is the ballot sent to the individual voter. But even that is futile in many cases. And there is no way at all to account for the fate of a ballot that is mailed out but not returned.
It is easy enough to verify a voter's identify at a polling place as the election judge has a living, breathing person in front of them. There is also a good chance others will recognize the voter as a neighbor or friend.
But there is no way to verify the entity who fills out a mail ballot. That can even be done by machine (election officials are not the only ones with computers and scanners). Once registered the only check on the identity of a ballot returned by mail is a signature, which is often "verified" solely by a machine sold by the same vendor providing the electronic voting machine. Apparently no election official thinks anyone else might have a signature writing machine or scanners. Just have to be careful the signature on the envelope matches the one used when the voter registration form was mailed in. But computers are really good at tasks like that...
Legally you must physically live in the precinct where you vote and provide a physical residential address which you declare is your primary residence during the year. But because the post office does not deliver mail to all addresses, voters may also list a mailing address, e.g., a post office box, where they have their mail sent. And anyone who has walked a precinct for a candidate or issue knows that some of the addresses given by voters simply don't exist, or are businesses or warehouses.
While the percentage of citizens who moved in 2010 is down to about 12% from previous records, that is still millions of voters who have moved out of their precinct. A very tiny fraction of these individuals notify the county clerk they moved. If they move somewhere else in the same state, and register to vote at their new address, a voter registration database with well-designed, intact referential integrity will catch that and enforce the change. But in my years as a database architect and modeler I rarely saw a database with intact referential integrity and "security by obscurity" prevents any such detailed examination of a state's voter registration database. Also, if the voter moves to another state it is virtually certain the two states won't be comparing voter registrations. Thus, thousands have been verified as voting in New York and Florida in the same election.
Mail ballots are certainly not publicly counted by citizen election judges. The usual procedure today is to use optical scanners located in the proverbial "back room" that are connected to a central server (computer). Since that server controls the votes cast for all candidates and issues in the county why not simply hack that computer and have the vote tally come out the way you want it to? That way you control the votes of everyone in the county rather than exerting the effort and risk of dealing with a few hundred physical ballots.
Recall that elections are run by clerks, who have taken the word of the voting machine vendor that their systems are accurate, calibrated, certified, and have never been known to fail or produce spurious and inaccurate results. Further, because the clerk usually lacks in-house expertise they have a maintenance agreement with the vendor to maintain and operate the voting machines. Anyone else see some potential problems here?
Lets start with the fact that ofttimes during an election the vendors simply hire technicians off the street and give them enough training on the equipment that they look like experts next to the computer-challenged clerks. Rarely are background checks run on these technicians. Why should they be when companies like Diebold (of the many names), ES&S, etc., are known to have convicted felons working as executives.
Mail ballots add to the risk of network hacking as the hacker receives a copy of the target ballot weeks in advance of the election. So they know well in advance what candidates and issues will appear on the official ballot as well as their positions on the ballot.
(2) A proximate threat. Many computers use wireless keyboards, a mouse, or some other WiFi connections. Anyone within range, sometimes up to several hundred yards, can hack into the system via the wireless connections. In fact, early versions of the Diebold AccuVote system had an IR port so that technicians could communicate with the scanner via their cell phone. Such "back doors" probably, almost certainly still exist.
Any communication by any means, wireless, wired, network, CD, DVD, flash drive, etc. between an infected computer and the server used to count mail ballots will probably result in the infection of the proximate computer.
(3) An insider threat. Most cases where election fraud has been proven and convictions resulted have involved insiders, often the county clerk. I've documented a number of such cases but very frequently it is impossible to distinguish between deliberate fraud and simple incompetence. And obviously the county clerk points the finger at the voting system vendor and vice versa.
Clearly it is easier to arrest and convict local officials than a hacker sitting in China, so there is a statistical bias when reviewing who is actually doing the hacking. But anyone who has access to the voting machines at any time, for any reason, has the potential to corrupt or infect the server. And with mail ballots it isn't necessary to infect the hundreds of machines used in polling place elections. Mail ballots are inevitably counted on one, or at most a few servers at a central location. So anyone with a flash drive and a few minutes of access to the server can hack it.
Nor does insider hacking need to be deliberate. All servers/computers need to be upgraded, repaired, have software installed, be connected to new hardware devices, tested, reprogrammed, and on and on. During any of these operations the technician may inadvertently insert malware that can later be used to control or hack the election. Once the server is infected, and the hacker receives a mail ballot so they know what the target looks like, controlling the election is well within the capability of many teenage hackers today.
(4) Compromised hardware or software. Virtually all of the motherboards for voting machines are Made in China. While the home offices may be in Taiwan or Hong Kong almost all the factory work is done in mainland China. That is also true of the computer "chips," i.e. large scale integrated circuits (LSI), that go on to the motherboards. And typically the LSI's are pre-programmed with firmware programs.
So China is both manufacturing and doing the basic programming of all our electronic voting machines. And testing for embedded malware is expensive and, even if recognized, difficult to remove and reprogram the firmware without which the chip typically won't function. I suggest it is safe to assume the hardware in these machines is compromised.
Then there is the problem of software. Most computer voting machines use some version of Windows CE operating system (OS), which is often installed in firmware in read-only memory (ROM) on the motherboard. It shouldn't be necessary to elaborate on all the viruses, rootkits, and the multitude of other malware associated with computers running Windows OS, and the vulnerability of any computer running Windows.
Bev Harris, Avi Rubin, and many others have examined the multitude of problems associated with the software used in electronic voting machines. Probably the best current review is the book Broken Ballots by Doug Jones and Barbara Simons. Suffice it to say that all experts who have had experience with this software have been, shall we say, appalled at the basic lack of security in these systems.
The above threats, of course, apply to all computers but mail ballots dramatically increase the ease of hacking the server used to count these ballots. Only one, or at most a few servers need to be hacked rather than hundreds or thousands as with, say, DREs. Two, the hacker will have the target ballot in their possession long before the election. Three, there is enormous incentive to hack these machines in order to control an election outcome. Four, since counting of mail ballots begins days and sometimes weeks before Election Day hacking can provide valuable data on what the early voting is showing.
In brief, given the incentives it should be assumed that the server being used in a mail ballot election has been hacked. Nor can it be safely assumed that available methods of detecting intrusions and malware are adequate to protect the server.
For all intents and purposes mail ballot elections eliminate party and public poll watchers. At best an observer can stand and watch ballots be fed into the scanner, but often today such watchers are placed behind a window and cannot actually enter the computer room.
In some cases watchers are allowed to see how ballots rejected by the scanner are copied on to new ballots. But it is extremely rare that poll watchers can directly observe how mail ballots are opened, the signature verified, batches of ballots assembled and controlled, counted, etc.
Then there is the simple issue that mail ballots are counted over days and weeks and very few people have the time or stamina to simply stand and watch ballots being fed into a machine for such an extended period.
Obviously with mail ballots they cannot, or at least are not counted in public at the precinct. Nor are they stored in the same ballot box as ballots cast at the precinct. That is an open invitation to errors and fraud.
In most cases mail ballots are counted in the proverbial "back room" where the public is not allowed for "security" reasons. Worse, in many cases, e.g., Denver, we have found the ballots being "counted" solely by election officials or the vendor's representative. Of course the counting is being done on the vendor's equipment and both their representative and the county clerk have great incentives to cover up any mistakes or errors that occur.
As a last ditch check on the validity and accuracy of an election, most jurisdictions have a canvass board composed of the county clerk or their representative(s), one or two representatives from each major party, and perhaps another member or two.
With mail ballot elections, or even mixed elections where all ballots are counted on electronic voting machines at polling places and an electronic poll book is used, a canvass board is reduced to looking at the numbers generated by the computers and simply putting their imprimatur on the computer printout of the poll book and election results.
Some audits, including hand counts of ballots from some races in some precincts are mandated in some areas. But current audit practices have been proven to be statistically meaningless. Further, county clerks have often cherry picked the precincts to audit.
The multitude of additional problems introduced by UOCAVA, such as sending ballots by email, are beyond the scope of this essay.
Even more frightening are the continued attempts to introduce Internet voting into public elections.An introduction to these problems are presented in an article by Jefferson, Rubin, Simons, and Wagner.
Even this rather surficial review of the problems of mail ballot elections makes it clear that virtually every protection developed over centuries to ensure and enforce honest elections has been compromised.
In addition to the push provided by the move to electronic voting machines, election officials and others used arguments of cost savings, voter turnout, and convenience to justify moving to "no excuse" mail ballots and eventually mail ballot elections.
Cost savings were generally based on the fact that ballots were only mailed to "active" voters, typically around 50-60% of the registered voters. However, with the move to mail ballot elections ballots are mailed to all registered voters which has greatly increased postal expenses as the clerks must now pay return postage as well on all returned ballots, which cannot be forwarded. From available information 10-15% of mailed ballots are usually returned by the post office. Also, election officials have been forced to provide ballot boxes where citizens can directly deposit their ballots when they don't trust the postal service. All in all the costs between a polling place election and a mail ballot election may be small.
Many jurisdictions have moved to mixed elections where there is in person early voting for a week or more before Election Day, polling places on Election Day, as well as mailing ballots to a substantial fraction of registered voters. Such mixed elections are definitely more expensive than either a polling place or mail ballot election. So with cost comparisons it is important to ensure one is comparing apples with apples, a rare event with bureaucracies.
Computer voting was also promoted as being able to provide virtually instantaneous election results for a news hungry press. With mail ballots most of the ballots would supposedly be counted prior to election day. That promise has not been upheld either, what with numerous machine breakdowns, too few DREs in many precincts, delays in the mail, hurricanes, and the general operation of Murphy's Law.
Voter turnout is generally acknowledged to have initially increased with mail ballots. But with time turnout appears to sink back to previous levels. So there is apparently no consistent voter turnout advantage with mail ballot elections.